Machine vision detection of bonemeal in animal feed samples

Non-invasive sensing for food reassurance

The main food quality traits of interest using non-invasive sensing techniques are sensory characteristics, chemical composition, physicochemical properties, health-protecting properties, nutritional characteristics and safety. A wide range of non-invasive sensing techniques, from optical, acoustical, electrical, to nuclear magnetic, X-ray, biosensor, microwave and terahertz, are organized according to physical principle.

Importance of spatial and spectral data reduction in the detection of internal defects in food products

Despite the importance of data reduction as part of the processing of reflection-based classifications, this study represents one of the first in which the effects of both spatial and spectral data reductions on classification accuracies are quantified. Furthermore, the effects of approaches to data reduction were quantified for two separate classification methods, linear discriminant analysis (LDA) and support vector machine (SVM). As the model dataset, reflection data were acquired using a hyperspectral camera in 230 spectral channels from 401 to 879 nm (spectral resolution of 2.1 nm) from field pea (Pisum sativum) samples with and without internal pea weevil (Bruchus pisorum) infestation. We deployed five levels of spatial data reduction (binning) and eight levels of spectral data reduction (40 datasets). Forward stepwise LDA was used to select and include only spectral channels contributing the most to the separation of pixels from non-infested and infested field peas. Classification accuracies obtained with LDA and SVM were based on the classification of independent validation datasets. Overall, SVMs had significantly higher classification accuracies than LDAs (P < 0.01). There was a negative association between pixel resolution and classification accuracy, while spectral binning equivalent to up to 98% data reduction had negligible effect on classification accuracies. This study supports the potential use of reflection-based technologies in the quality control of food products with internal defects, and it highlights that spatial and spectral data reductions can (1) improve classification accuracies, (2) vastly decrease computer constraints, and (3) reduce analytical concerns associated with classifications of large and high-dimensional datasets.

How do "mute" cicadas produce their calling songs?

Insects have evolved a variety of structures and mechanisms to produce sounds, which are used for communication both within and between species. Among acoustic insects, cicada males are particularly known for their loud and diverse sounds which function importantly in communication. The main method of sound production in cicadas is the tymbal mechanism, and a relative small number of cicada species possess both tymbal and stridulatory organs. However, cicadas of the genus Karenia do not have any specialized sound-producing structures, so they are referred to as "mute". This denomination is quite misleading, as they indeed produce sounds. Here, we investigate the sound-producing mechanism and acoustic communication of the "mute" cicada, Karenia caelatata, and discover a new sound-production mechanism for cicadas: i.e., K. caelatata produces impact sounds by banging the forewing costa against the operculum. The temporal, frequency and amplitude characteristics of the impact sounds are described. Morphological studies and reflectance-based analyses reveal that the structures involved in sound production of K. caelatata (i.e., forewing, operculum, cruciform elevation, and wing-holding groove on scutellum) are all morphologically modified. Acoustic playback experiments and behavioral observations suggest that the impact sounds of K. caelatata are used in intraspecific communication and function as calling songs. The new sound-production mechanism expands our knowledge on the diversity of acoustic signaling behavior in cicadas and further underscores the need for more bioacoustic studies on cicadas which lack tymbal mechanism.

Rapid localization of bone fragments on surfaces using back-projection and hyperspectral imaging

Manual localization of bone fragments on the ground or on complex surfaces in relation to accidents or criminal activity may be time-consuming and challenging. It is here investigated whether combining a near-infrared hyperspectral camera and chemometric modeling with false color back-projection can be used for rapid localization of bone fragments. The approach is noninvasive and highlights the spatial distribution of various compounds/properties to facilitate manual inspection of surfaces. Discriminant partial least squares regression is used to classify between bone and nonbone spectra from the hyperspectral camera. A predictive model (>95% prediction ability) is constructed from raw chicken bones mixed with stone, sand, leaves, moss, and wood. The model uses features in the near-infrared spectrum which may be selective for bones in general and is able to identify a wide variety of bones from different animals and contexts, including aged and weathered bone.

Reflectance-based identification of parasitized host eggs and adult Trichogramma specimens

A wide range of imaging and spectroscopy technologies is used in medical diagnostics, quality control in production systems, military applications, stress detection in agriculture, and ecological studies of both terrestrial and aquatic organisms. In this study, we hypothesized that reflectance profiling can be used to successfully classify animals that are otherwise very challenging to classify. We acquired hyperspectral images from adult specimens of the egg parasitoid genus Trichogramma (T. galloi, T. pretiosum and T. atopovirilia), which are ~1.0 mm in length. We also acquired hyperspectral images from host eggs containing developing Trichogramma instar and pupae. These obligate egg endoparasitoid species are commercially available as natural enemies of lepidopteran pests in food production systems. Because of their minute size and physical resemblance, classification is time consuming and requires a high level of technical experience. The classification of reflectance profiles was based on a combination of average reflectance and variogram parameters (describing the spatial structure of reflectance data) of reflectance values in individual spectral bands. Although variogram parameters (variogram analysis) are commonly used in large-scale spatial research (i.e. geoscience and landscape ecology), they have only recently been used in classification of high-resolution hyperspectral imaging data. The classification model of parasitized host eggs was equally successful for each of the three species and was successfully validated with independent data sets (>90% classification accuracy). The classification model of adult specimens accurately separated T. atopovirilia from the other two species, but specimens of T. galloi and T. pretiosum could not be accurately separated. Interestingly, molecular-based classification (using the DNA sequence of the internally transcribed spacer ITS2) of Trichogramma species published elsewhere corroborates the classification, as T. galloi and T. pretiosum are closely related and comparatively distant from T. atopovirilia. Our results emphasize the importance of using high-spectral and high-spatial resolution data in the classification of organism relatedness, and hyperspectral imaging may be of relevance to a wide range of commercial (i.e. producers of biocontrol agents), taxonomic and evolutionary research applications.

Application of hyperspectral imaging in food safety inspection and control: a review

Food safety is a great public concern, and outbreaks of food-borne illnesses can lead to disturbance to the society. Consequently, fast and nondestructive methods are required for sensing the safety situation of produce. As an emerging technology, hyperspectral imaging has been successfully employed in food safety inspection and control. After presenting the fundamentals of hyperspectral imaging, this paper provides a comprehensive review on its application in determination of physical, chemical, and biological contamination on food products. Additionally, other studies, including detecting meat and meat bone in feedstuffs as well as organic residue on food processing equipment, are also reported due to their close relationship with food safety control. With these applications, it can be demonstrated that miscellaneous hyperspectral imaging techniques including near-infrared hyperspectral imaging, fluorescence hyperspectral imaging, and Raman hyperspectral imaging or their combinations are powerful tools for food safety surveillance. Moreover, it is envisaged that hyperspectral imaging can be considered as an alternative technique for conventional methods in realizing inspection automation, leading to the elimination of the occurrence of food safety problems at the utmost.

Use of weighting algorithms to improve traditional support vector machine based classifications of reflectance data

Support vector machine (SVM) is widely used in classification of hyperspectral reflectance data. In traditional SVM, features are generated from all or subsets of spectral bands with each feature contributing equally to the classification. In classification of small hyperspectral reflectance data sets, a common challenge is Hughes phenomenon, which is caused by many redundant features and resulting in subsequent poor classification accuracy. In this study, we examined two approaches to assigning weights to SVM features to increase classification accuracy and reduce adverse effects of Hughes phenomenon: 1) "RSVM" refers to support vector machine with relief feature weighting algorithm, and 2) "FRSVM" refers to support vector machine with fuzzy relief feature weighting algorithm. We used standardized weights to extract a subset of features with high classification contribution. Analyses were conducted on a reflectance data set of individual corn kernels from three inbred lines and a public data set with three selected land-cover classes. Both weighting methods and reduction of features increased classification accuracy of traditional SVM and therefore reduced adverse effects of Hughes phenomenon.